Open Research Online
The Open University’s repository of research publications
and other research outputs
The attention-grammar interface: eye-gaze cues
structural choice in children and adults
Journal Article
How to cite:
Ibbotson, Paul; Elena, Lieven and MIchael, Tomasello (2013). The attention-grammar interface: eyegaze cues structural choice in children and adults. Cognitive Linguistics, 24(3) pp. 457–481.
For guidance on citations see FAQs.
c 2013 Walter de Gruyter
Version: Accepted Manuscript
Link(s) to article on publisher’s website:
http://dx.doi.org/doi:10.1515/cog-2013-0020
Copyright and Moral Rights for the articles on this site are retained by the individual authors and/or other
copyright owners. For more information on Open Research Online’s data policy on reuse of materials please
consult the policies page.
oro.open.ac.uk
The Attention-Grammar Interface
Running Head: THE ATTENTION-GRAMMAR INTERFACE.
The Attention-Grammar Interface:
Eye-Gaze Cues Structural Choice in Children and Adults
1
The Attention-Grammar Interface
Abstract
We investigated whether children (3- and 4-year-olds) and adults can use the active passive
alternation – essentially a choice of subject– in a way that is consistent with the eye-gaze of
the speaker. Previous work suggests the function of the subject position can be grounded in
attentional mechanisms (Tomlin, 1995; 1997). Eye-gaze is one powerful source of directing
attention that we know adults and young children are sensitive to; furthermore, we know
adults are more likely to look at the subject of their sentence than any other character
(Gleitman, January, Nappa & Trueswell, 2007; Griffin & Bock, 2000). We demonstrate that
older children and adults are able to use speaker-gaze to choose a felicitous subject when
describing a scene with both agent-focused and patient focused cues. Integrating attentional
and grammatical information in this way allows children to limit the degrees of freedom on
what the function of certain linguistic constructions might be.
Key words
Attention-Grammar Interface, Eye-Gaze, Subject, Argument-Structure Constructions, Social
Cognition.
2
The Attention-Grammar Interface
Introduction
Understanding the interface between attention and grammar is a central concern in many
functional linguistic theories (e.g., Langacker, 1991; Talmy, 2007) and an important topic in
psycholinguistic research (e.g., Henderson and Ferreira, 2004; Trueswell and Tanenhaus,
2005). Directing attention via referential and perceptual priming causes people to construe a
scene in a particular way and typically this is reflected in the linguistic structures people use
(Prentice, 1967; Turner and Rommetveit, 1968; Posner, 1980; Tomlin 1995, 1997; see also
review by Myachykov et al., 2011). Here we investigate the extent to which eye-gaze
influences speakers’ construal of a scene in both adults and children. We offer a
developmental perspective as children have to learn the particulars of the form-construal
mapping in their language and using social-cognitive information such as eye-gaze might be
one cue (among many others) as to how this works. First, we introduce the cognitive
linguistic perspective on construal and the attention-grammar interface. Second, we discuss
the hypothesis that the grammatical notion of subject can be grounded in the cognitive
concept of attention. Third, we outline the development of subject knowledge, attention and
eye-gaze sensitivity. Finally, we set out how this previous work relates the hypothesis that
social-cognitive cues such as eye-gaze might scaffold construction learning.
Construal and the attention-grammar interface
The same state of affairs in the world can be encoded by the use of different linguistic
devices to communicate a nuanced range of perspectives. For example:
1a. the roof slopes gently downwards
1b. the roof slopes gently upwards
3
The Attention-Grammar Interface
The two scenes referring to the same roof can be mentally viewed or ‘construed’ from either
above (1a) or below (1b) (cf. Langacker 1988: 62). Languages have of course evolved many
different ways to alter how a particular concept is construed in the mind including different
structural frames:
2a. the dog chased the cat
2b. the cat was chased by the dog
Note that the scenes in 1a-1b and 2a-2b are truth-conditionally equivalent in the sense that
the state of affairs in the world which requires the statements to be true is the same for both.
For example, in 1a and 1b there is a roof that exists such that it is angled at x degrees. The
different expressions are therefore not describing different facts about the world rather they
are conventionalized and prefabricated ways of expressing different perspectives. In some
functional linguistic approaches (e.g., Goldberg, 2006) more abstract forms such as
argument-structure constructions carry their own meaning independent of the items that
appear in them; thus the hearer’ s attention is still guided to what happened to the gazzer in
the passive ‘the gazzer was mibbed by the pubber’ regardless of the fact that we do not
know what the words mean. Contrast this with theoretical frameworks that analyse 2b as the
movement of the cat from the post verbal position into subject position (e.g., Radford et al.,
1999). In the functionalist approach, “surface grammatical form does not conceal a ‘truer’,
deeper level of grammatical organization; rather, it itself embodies the conventional means a
language employs for the structuring and symbolization of semantic content” (Langacker
1987: 46-7).
Some of the most extensively investigated attention-directing devices are those
components of dynamic events, including those that highlight the conceptual distinctions of
4
The Attention-Grammar Interface
containment/support, path/manner, source/goal and figure/ground (Choi and Bowerman,
1991; Jackendoff, 1983; Lakoff, 1987; Talmy, 1985, 2000, 2007). The notion of
figure/ground – whereby some information is highlighted with respect to relatively stable
background– has been particularly well studied in relation to the English subject (for a
similar conceptual division see Fillmore’s (1976) "frame/highlighting" and Langacker’s
(1987) "base/profiling")
The English Subject and Attention
The subject of an English1 clause is a ‘mosaic’ of prototypical coding and behavioural
features (Keenan, 1976). For example, the subject typically comes before the verb and
triggers agreement with it (e.g., She smiles, not She smile), has a special pronominal form
(e.g., She smiles, not Her smiles) and entails certain structural properties (e.g., only the
subject can leave in She smiled at him and left). The subject position is also associated with
certain discourse properties2; most relevantly for this study, the foregrounding of items that
appear in that position and, by definition, backgrounding other items in the clause
(depending on the choice of linguistic terminology this function is also been variously
referred to as figure-ground, perspective, theme, aboutness and prominence e.g., Talmy,
1985; MacWhinney, 1977; Langacker, 1991). Tomlin has shown that the foregrounding
function of subject can essentially be re-described in terms of the cognitive concept of
attention (1995, 1997). The key idea here is that attentional mechanisms privilege some
forms of information over others by ‘gating’ perceptual input, sustaining focus on what is
1
Cross-linguistically speaking, there have been about 30 different grammatical features that
have been variously attributed to the concept of “subject” (for example controlling verb
agreement, determining the actor in a subjectless second coordinate clause and so on) and
any one language ‘subject’ is only a subset of these features which do not necessarily
overlap.
2
For example, the subject position is associated with given information whereas the object
position is associated with new information (e.g., Halliday, 1985).
5
The Attention-Grammar Interface
foregrounded. What is foregrounded then becomes the subject of the sentence (see Figure 1).
This is most clearly exemplified by the active passive alternation in English, the function of
which allows speakers to focus attention on what the agent did (active) in contrast to what
happened to the patient (passive), examples 2a-2b and Figure 1. The utility of defocusing
the agent’s role in an action or state of affairs has not escaped many politicians, most
famously realized in the non-apology apology “...mistakes were made…”
Figure 1. A schematic of the relationship between the subject role and the active passive
voice alternation. Participants that appear in the subject role receive attention. Active
sentences are the unmarked form and are relatively more frequent transitive sentences than
passives, as indicated by the heavier outline. Arrows indicate the direction of transitivity.
To test the hypothesis that subject-is-theme-is-attention Tomlin (1995) asked participants to
watch prototypical transitive scenes of two fish approaching each other until one swallows
the other and swims away. Tomlin manipulated the attention of participants by placing a
6
The Attention-Grammar Interface
flashing red arrow above one or other of the fishes 75 ms before the eating action was
completed. Participants were asked to keep their eyes on the character the arrow pointed at
and describe what they saw. The majority of adult speakers performed as the subject-istheme-is-attention hypothesis predicted; on the cued agent animations (the arrow was above
the fish that was doing the eating) the agent was assigned the subject position, and the
clause was active. On the cued patient animations (the arrow was above the fish being eaten)
the patient was assigned subject position and the overall clause was passive. Gleitman,
January, Nappa & Trueswell, (2007) obtained similar results even when the cue was implicit
(participants were largely unaware of the cue because it appeared so briefly), although with
a decreased effect size of passivisation.
Items that appear at the start of an utterance occupy a salient slot and thus could
trigger structural organization somewhat independently of grammatical status (MacWhinney,
1977). Because the sentence initial position is confounded with Subject role in English
transitive sentences it is difficult to differentiate between a linear-ordering versus a
grammatical-role account of the priming effects. To do so we must turn to languages that
permit, under certain pragmatic contexts, more flexible word orders. Three recent studies
analyzed perceptually primed structural choice in Russian (Myachykov and Tomlin, 2008),
Finnish (Myachykov et al. 2010), and Korean (Hwang and Kaiser 2009). Overall the studies
suggested that in flexible word-order languages the extent of perceptual priming is
consistently weaker than in the fixed word-order languages. Myachykov et al (2010)
propose that speakers universally attempt to employ the grammatical-role assignment
mechanism in order to represent the perceptually salient referent but this interacts in
complex ways with the availability and reliability of the linguistic resources of the particular
language. In languages like Russian and Finnish, for example, passives are rare or largely
7
The Attention-Grammar Interface
dispreferred and as a result, a linear-ordering mechanism is used to accommodate referential
salience in terms of word order.
The exact nature of the attention-grammar interface is still uncertain and of course,
subject position is one attentional cue among many others. For example, in unmarked cases,
English tends to correlate theme with given information and subject position, and focus with
new information and object position (and usually also prosodic stress). The famous Moses
illusion takes advantage of this pattern: when asked “How many animals of each kind did
Moses take on the Ark,” most people respond “two,” even though it can be independently
established that they know that it was Noah, not Moses, who took the animals on the Ark
(Erickson & Mattson, 1981). The fundamental role of attention in this process is underlined
by the fact the illusion can be ameliorated when attention is focused on the incongruent item
using structures such as clefts (3a) and there-insertions (3b) (Traxler, 2012)
3a. It was Moses who took two of each kind of animal on the Ark.
3b. There was a guy called Moses who took two of each kind of animal on the Ark.
While acknowledging the role that focus and other salience cues play at the attentiongrammar interface, we concentrate here on the subject position for the theoretical reasons
outlined above. Recognizing that these and other factors do influence structural choice, the
counterbalancing procedure controls for these effects, see Design.
The development of subject knowledge, attention and eye-gaze sensitivity
The overall developmental picture is that children acquire the different features of
the English subject gradually and at different times in a ‘mosaic’ fashion (Rispoli, 1991).
Young children identify ‘subject’ as the most animate participant, or the first-mentioned
8
The Attention-Grammar Interface
participant, or the agent, which suggests something less abstract than an adult-like notion of
subject (see Corrigan, 1998 for a review). Experimentally Braine et al. (1993) have shown
that mastery over the notion of English subject appears at around 5-6 years of age (see also
Corrigan, 1998). Consistent with this, typically children do not produce full passives in
spontaneous speech until about 4-5 years-of-age however performance can be significantly
boosted when the passive form is supported with case marked pronouns (Ibbotson et al.,
2009) with training (Pinker, Lebeaux & Frost 1987; Brooks & Tomasello, 1999) and when
the passive form is more frequent in the ambient language (relative to Indo-European
languages: Allen & Crago, 1996 for Inuktitut; Suzman, 1985 for Zulu; Pye & Poz, 1988 for
K’iche’ Mayan). Following Croft (2001) one explanation for this relatively late and
piecemeal acquisition pattern is that in reality, abstract constructions such as intransitive,
transitive, passive and there-constructions actually have their own subject. They may only
be united by analogy later on in development under something like a highly schematic
subject-predicate construction (Tomasello, 2003).
In terms of the development of attention, unsurprisingly, younger children are more
easily distracted by task-irrelevant events (Enns & Girgus, 1985; Well, Larch, & Anderson,
1980; Carlson, 2005), they search less efficiently for a specified target (Day, 1978; Gibson
& Yonas, 1966), they are less able to sustain attention on a given task (Corkum et al., 1995;
Kannass and Oakes, 2008) and they are less able to switch efficiently from task to task
(Guttentag, 1985; Pick & Frankel, 1974). The general developmental picture for infants is
one of increasing cognitive control so that by around their fourth year they have begun to
orientate their attention with volition and flexibility (Hughes, 1998; Ruff & Rothbart, 1996).
Orienting of attention in this context refers to the alignment of some internal mechanism
9
The Attention-Grammar Interface
with an external sensory input source that results in the preferential processing of that input.
The external sensory input source we are most interested in here is eye-gaze.
Humans show a strong sensitivity to eye-gaze from birth (Farroni, Csibra, Simion, &
Johnson, 2002). Neonates can follow gaze if the pupils are seen to move (Farroni,
Massaccesi, et al., 2004) and at around 5-months-of-age they can discriminate between very
small horizontal deviations (5°) of eye gaze (Symons et al., 1998). Like adults, infants
process facial features in deeper way when gaze is directed towards them as compared with
averted gaze (Farroni et al., 2002; Farroni, Johnson, & Csibra, 2004; Farroni, Massaccesi,
Menon, & Johnson, 2007; Hood et al., 2003). Clearly, the capacity to use another person’s
eye gaze as a cue to attention develops very early in life, however, to begin with this might
be achieved with rather low-level, low-mentalising mechanisms, for example, the perceptual
geometry and luminance of the eye (Ando, 2002). Compared with other primates, humans
have a relatively large white sclera surrounding a small dark pupil and iris making eye-gaze
discrimination relatively easier in humans than in other animals (Kobayashi & Kohshima,
1997). Supporting the low-mentalising interpretation of eye-gaze sensitivity, a wide range of
species have a very accurate ability to determine whether they are being looked at (e.g.,
Burger, Gochfeld, & Murray, 1992; Burghardt & Greene, 1990; Perrett & Mistlin, 1991)
and nonhuman primates such as adult rhesus monkeys can discriminate between
photographs depicting direct gaze and gaze averted by 5°, the same ability that has been
reported in human infants (Campbell, Heywood, Cowey, Regard, & Landis, 1990; Symons,
Hains, & Muir, 1998).
Although infants’ eye-gaze sensitivities may be based on relatively simple
mechanisms (gaze perception), young children soon begin to integrate eye-gaze information
into a more sophisticated picture of how other people work including their future intentions
10
The Attention-Grammar Interface
and mental states (Baron-Cohen, 1994; see Striano & Reid, 2006, for a recent review). The
development of joint attention at 20 months predicts theory of mind abilities at 44 months
(Charman et al., 2001) underlining eye-gaze as a key component in the development of
social cognition in early life. Baron-Cohen, Campbell, Karmiloff-Smith, Grant, and Walker
(1995) found that children ages 3 and 4 years old deduce the direction of gaze of a
schematic face and they can ascribe mental states such as desires on the basis of the
direction of gaze (see also Lee, Eskritt, Symons, & Muir, 1998). Thus, understanding that
direction of gaze can indicate which objects a person knows exists, is currently attending to,
and holds a mental state about can help a child infer much about the current visual world
(although this understanding may not be as flexible as adults when cues conflict e.g., Friere,
Eskritt, & Lee, 2004; Pellicano & Rhodes, 2003; Pellicano and Rhodes, 2003).
Eye-gaze following at 6 months has been shown to correlate with vocabulary size at
18 months (Morales et al., 2000; Morales, Mundy, & Rojas, 1998) and in noun learning,
children can use eye-gaze, head posture and gesture to infer speakers’ referential intention
(e.g., Baldwin, 1991; Carpenter , Akhtar & Tomasello, 1998; Gergely, Bekkering & Király,
2002; Woodward & Sommerville, 2000). Nappa, Wessel, McEldoon, Gleitman &
Trueswell (2009) showed that 3-, 4- and 5-year-olds used the eye-gaze of the speaker to
infer the meaning of novel relational verbs (of the type chase vs. flee) in linguistically
uninformative contexts (e.g., He’s mooping him). Thus children who saw a speaker looking
at the chaser when they uttered the novel verb were more likely to attribute ‘chase-like’
semantics to the novel verb. The opposite effect was found when a speaker looked at the
flee-er.
Change in gaze direction is one of several behavioral cues that individuals use in
combination with changes in facial and vocal displays and body posture to mark the
11
The Attention-Grammar Interface
intention to act on an object (Mumme et al., 2007). Crucially, just prior to speaking, adults
are more likely to look at the subject of their sentence than any other character (Gleitman,
January, Nappa & Trueswell, 2007; Griffin & Bock, 2000). This raises the possibility that
children could use this cue in the input, probabilistically at least, to build a correspondence
between the perspective of an event and how that perspective is expressed in their language
(cf. Nappa et al., 2009). Just as verbs such as chase and flee can lead to different construals
of the same (perceptual) scene so can argument-structure constructions like the activepassive alternation, which are basically perspective-taking devices.
In summary, the above evidence raises the possibility that young children could use
the social-cognitive cue of eye-gaze – which directs attention – to infer the function of
grammatical subject – which is grounded in attention. We investigated this by exploring if
they can use the active passive alternation (essentially a choice of subject) in a way that is
consistent with the eye-gaze of the speaker. The developmental hypothesis is that if the
function of subject position is grounded in attentional mechanisms (Tomlin, 1995; 1997),
then we would expect that developing attentional abilities should interact with developing
linguistic ability to assign a subject. Thus we would expect different age groups to perform
differently. Our second hypothesis is more of a general theoretical point and is relevant to
adults and children. As the evidence we have reviewed suggests, some cognitive linguistic
frameworks suggest a close relationship between attention and linguistic performance (e.g.,
Givon 1992; Landau and Jackendoff 1993; MacWhinney 1977; Osgood and Bock 1977;
Talmy 2007). This general assumption leads to the specific hypothesis that a speaker’s
attention can be guided by the eye-gaze of their interlocutor and that this in turn affects the
structural choice of their utterances, specifically whether to use an active or passive sentence.
12
The Attention-Grammar Interface
Methods
Participants
A total of 91 participants were tested. For each age group we report the numbers in each
group (N) the mean age in years (M) and the standard deviation of the group as calculated
on months-of-age (SD). For the Test condition: 3-year-olds (N=21, M=3.55, SD±3.66), 4year-olds (N=23, M=4.39, SD±2.79) and adults (N=11, M=23.31, SD±87.60). For the
Control Condition: 3-year-olds (N=13, M=3.45, SD±3.63), 4-year-olds (N=12, M=4.38,
SD±3.60) and adults (N=11, M=19.08, SD±41.25). 7 children were excluded from the
analysis because of fussiness (4) or not producing a novel verb (3) – for example they
persisted with “the dog’s hitting the cat” even after correction, instead of the intended “the
dog’s tamming the cat”.
Design and Materials
The between-subjects factors were Age (3-year-old, 4-year-old and adults) and Condition
(Test, Control) and the within-subjects factor for the Test Condition was Gaze (Agent-Focus,
Patient-Focus). Ideally, to increase the power we would have treated each child as its own
control with a within-subjects design. We assumed that most children and adults would
provide answers in the experiment with an active voice if not cued to do otherwise (as
indeed they did). We reasoned that if we ran a cued and non-cued trial with the same
individual there was a risk that we would structurally prime the active response to such a
degree that the already weakly represented passive would not get produced at all. We were
interested in the degree to which participants could switch between active and passive
responses in a way that was congruent with eye-gaze. By the time a participant answered
with 8 actives responses (as they most likely would have done in the test condition) it would
13
The Attention-Grammar Interface
be much harder to elicit a passive than if they have just finished our training phase (see
below). Likewise, running the control after the test might have artificially boosted passive
responses in the control condition. Means and standard deviations of age between Test and
Control were comparable (see Participants). Increasing the activation levels/ease of retrieval
of the either active or passive was avoided by keeping control and test participants separate.
Although this reduced the power, as it turned out the effect was still strong enough to detect.
If participants came to the experiment with no preference for describing a scene with
either actives or passives and responded at random then we could expect the proportion of
congruent answers to be 50%, that is, an active response to agent-focused question and a
passive response to a patient-focused question. If participants came to the experiment with a
preference for describing a scene with only actives then again we could expect the
proportion of congruent answers to be 50% – participants score 4/4 on the active-focused
questions but 0/4 on passive focused questions. The design therefore takes account of
baseline preference with respect to congruency. Despite this methodical treatment of
participant’s preference we still ran a control to check there was nothing strange about the
puppets used or experimental set-up and to measure the different age groups’ ‘default’
response that might vary for whatever reason. We therefore present the control data
alongside the test data in the results section as it impacts on how easily participants can
switch from their ‘default’ response to an eye-gaze cued response.
There were 8 test questions per participant, with an equal number of Agent-Focused
and Patient-Focused trials, the orders of which were randomised. Over the experiment, the
order of appearance of model sentences (training phase, see below), the left/right appearance
of the participant in focus and the participant acting as agent were counterbalanced. There
were six hand puppets used, randomly assigned to either the training phase or the test phase.
14
The Attention-Grammar Interface
Referential salience and conceptual accessibility affect the degree to which a
participant might attend to a particular character. Care was taken to match the puppets on
size, general colourfulness and animals that could be considered prototypical instances of
their own kind, (a monkey, a pig, a horse, a dog, a cat, and a bear). All actions were
reversible transitives and the experimenter took care to make the actions as comparable as
possible between trials (as far is possible with the act-out method). Importantly however, the
counterbalancing procedure controls for salience. For example, if participants considered the
pig more salient than the cat and correspondingly chose this as the subject, the pig appeared
both as the focus of attention in active and passive conditions, and the unfocused participant
of active and passive conditions over the course of the experiment.
Procedure
Training Phase
The experimenter modelled novel verb, (tamming) using animal hand puppets. This was
done twice in the active voice (e.g., “look! the horse is tamming the goat”) and twice in the
passive voice (e.g., “look! the frog is getting tammed by the cow !”). After each model the
experimenter asked the child or adult to repeat the sentence (e.g., “can you say the horse is
tamming the goat?”). The experiment did not progress until the participants had correctly
repeated all four of the model sentences at least once. The action consisted of an agent
hitting the top of a patient’s head with a cardboard tube (see Figure 1). We used the
cardboard tube to increase the distance between participants and thus make it easier to
determine whether the experimenter was looking at the agent or patient (yet still remain a
causal transitive scene). Throughout the training phase, the experimenter did not use eyegaze to direct attention onto either of the characters.
15
The Attention-Grammar Interface
Test Phase
The training puppets were abandoned and four new hand puppets were introduced. The
experimenter asked the participant to name the animals, which all did correctly. The
experimenter then said “we’re going to play a different game now, this time you have to tell
me what the animals are doing”. The experimenter looked at the participant, engaged eye
contact, looked at the target animal (either the agent or patient of the impending action)
looked back at the participant and then finally looked back to the target animal again. Thus
the idea was to establish triadic joint-attention between speaker, addressee and referent. The
experimenter then performed the action while continuing to look at the target referent.
Shortly after the action had begun the experimenter asked, “what’s happening?” If the
participant did not respond, the experimenter persisted for a while with the same question
taking care not to linguistically cue them into an active or passive response (e.g., “what is
the cat doing”? or “what’s happening to the dog”?) – thus we tried to create an analogous
situation to the Nappa et al. ‘linguistically uninformative’ condition for verb learning (2009).
Children’s and adult’s responses were recorded and later coded for active or passive voice
and the order in which participants were mentioned. We also recorded whether the subject
of the sentence (either agent or patient) was congruent or incongruent with the eye-gaze of
the speaker/experimenter. For example, if the speaker was looking at the agent of the action
and the participant responded with the agent in subject position (i.e., active: agent-V-patient)
it was coded as congruent. If the participant responded with the patient in subject position
(i.e., passive: patient-Aux-V-(agent)) it was coded as incongruent. It is important to note
that either an active or a passive answer to the question “what’s happening” is equally
grammatical with respect to who is doing what to whom (assuming the arguments map onto
what is actually happening in the scene). We are interested in what could be considered the
16
The Attention-Grammar Interface
most pragmatically felicitous response (following adult performance in Tomlin, 1995; 1997
and anticipated adult performance in this study), namely those responses where the subject
of the utterance maps onto where the attention has been directed, in this case via the eyegaze of the speaker. Adults and children experienced the same experimental procedure with
the same materials. While the hand-puppet procedure might have appeared strange to some
adults it was necessary to make the results directly comparable – if we had not used hand
puppets any difference between adults and children could be attributable to the different
methodologies used.
The control condition contained a procedure that was identical to that described
above except for the fact that the experimenter did not attempt to focus the attention of the
child or adult onto any participant before the action began and looked straight ahead when
asking “what’s happening?” (Figure 2).
Agent-Focus
Control
Patient-Focus
Figure 2. In this example ‘the cat is tamming the dog’ or, depending on your perspective,
the ‘dog is getting tammed’. The eye-gaze of the experimenter cues attention towards either
the agent of the action (left), neither agent nor patient (centre) or the patient of the action
(right).
17
The Attention-Grammar Interface
Results
First we present a comparison between the Control Condition and the Test Condition. We
underline that the control condition was identical to the test condition except for the fact that
the experimenter did not attempt to focus the attention of the child or adult onto any
participant before the action began and looked straight ahead when asking “what’s
happening?” The results are presented in terms of Eye Gaze (Test Condition) and No Eye
Gaze (Control Condition) as a proportion of active responses, displayed in Figure 3 below.
Figure 3. Mean Proportion of Active Responses by Age Group.
18
The Attention-Grammar Interface
To determine whether there was a significant difference between Test and Control
conditions with respect to the mean proportion of active utterances we conducted a 2 (Eye
Gaze Focus, No Eye Gaze Focus) by 3 (3-Year-Olds, 4-Year-Olds, Adults) ANOVA with
mean proportion of active responses as the dependent variable. There was a significant
effect of Condition F(1,85) = 6.07, p=.016 ηp2 = .067, Age F(2,85) = 3.68, p=.029 ηp2 = .08
and no significant interaction between Age and Condition F(2,85) = 0.85, p=.428 ηp2 = .02.
This means Eye Gaze significantly affected the active/passive ratio for all age groups. What
is most relevant for the discussion is that the youngest age group show the strongest
preference for describing the scene (in the absence of eye-gaze focus) with active sentences.
This means the youngest age group has the strongest ‘default’ to overcome if they are to
switch to describing the scene with a passive.
We now present the analysis of the Test data with respect to the mean proportion of
congruent responses to eye-gaze (obviously we cannot analyse the control data in this way
as there is no eye gaze manipulation in the Control – note the change in y-axis from Figure 3
to 4). Figure 4 shows the mean proportion of congruent responses as a function of whether
the experimenter was looking at the agent of the action or the patient of the action for
different age groups.
19
The Attention-Grammar Interface
Figure 4. Mean Proportion of Congruent Responses as a function of Eye-Gaze and Age.
Note the change in y-axis from Figure 3.
Overall Congruency Comparisons
To determine the overall effect of eye gaze on congruent responses, we first conducted a 2
(Congruent, Incongruent) by 3 (3-Year-Old, 4-Year-Old, Adults) mixed model analysis of
variance (ANOVA). This showed a significant main effect of Congruency F(1,52) = 97.94,
p<0.001, ηp2 = .653 (M .737 > .263) and a significant interaction between Congruency and
Age F(2,52) = 14.36, p<0.001, ηp2 = .356. This means that overall there were more
20
The Attention-Grammar Interface
congruent responses than incongruent. The interaction suggests that the difference between
congruent and incongruent responses was not the same across all age groups. To analyze
this interaction further we conducted an ANOVA by Age. We found significant differences
between congruent and incongruent responses for 4-Year-Olds F(1, 22) = 36.59, p<0.001,
ηp2 = .625 and Adults F(1, 10) = 306.25, p<0.001 ηp2 = .968 but not for 3-Year-Olds F(1, 20)
= 3.50, p = 0.07 ηp2 = .149. This means that regardless of the type of eye-gaze focus
(agent/patient) only the 3-Year-Olds were unable to consistently match a congruent
response. We now analyze how this overall effect of eye-gaze on congruency is influenced
by whether the gaze is focused on the agent or patient.
Agent-Focused
We analyzed Agent-Focused trials with a 2 (Congruent, Incongruent) by 3 (3-Year-Old, 4Year-Old, Adults) mixed model analysis of variance (ANOVA). There was a significant
main effect of Congruency F(1, 52) = 52.49, p<0.001 ηp2 = .502 (M .817 > .183) and no
interaction between Congruency and Age F(2, 52) = 1.82, p=0.172 ηp2 = .066. This means
all age groups consistently produced active sentences when the experimenter was looking at
the agent.
Patient-Focused
Second, we analyzed Patient-Focused trials with a 2 (Congruent, Incongruent) by 3 (3-YearOld, 4-Year-Old, Adults) mixed model ANOVA. There was a significant main effect of
Congruency F(1, 52) = 9.25, p <0.001 ηp2 = .151 (M .657 > .343) and a significant
interaction between Congruency and Age F(2, 52) = 5.72, p<0.001 ηp2 = 180. The
interaction suggests that the difference between congruent and incongruent responses for
Patient-Focused trials was not the same across all age groups. To analyze this further we
21
The Attention-Grammar Interface
conducted an ANOVA by Age. We found significant differences between congruent and
incongruent responses for 4-Year-Olds F(1, 22) = 5.80, p=0.025, ηp2 = .209 and Adults F(1,
10) = 49.23, p<0.001 ηp2 = .831 but not for 3-Year-Olds F(1, 20) = 0.56, p=.460, ηp2 = .028.
This means only the 3-Year-Olds could not consistently produce passive sentences when the
experimenter was looking at the patient.
Finally we report that for both Control and Test conditions the vast majority of
passive responses from children and adults were truncated (e.g., “the cat got tammed”)
rather than full passives (e.g., “the cat got tammed by the dog”), Table 1.
Table 1. Proportion of Full Passives by Age Group and Condition (%)
3-Year-Olds
Test
Control
4/168 (2.38)
0/104 (0)
4-Year-Olds 11/184 (5.97)
Adults
2/96 (2.08)
19/88 (21.59) 12/88 (13.63)
This did not have a significant impact on the main analyses as we are primarily
interested in whether the (correct, i.e., congruent) agent/patient appeared in the subject role
as a function of speaker’s eye gaze. In principle there are other ways of describing the scene,
where the subject would be congruent with eye-gaze but the structure would not be active or
passive, such as Dog is hurting from cat tamming it. We did not witness any examples of
these types of sentences, perhaps because the training phase primed/reduced the degrees of
freedom on what participants thought is an appropriate response to the question “what’s
happening?”, i.e. an active or passive was modeled by the experimenter in the training phase
(and participants were asked to repeat them) and once participants had begun to answer with
these options they continued to do so. In sum, all ages gave more congruent responses than
22
The Attention-Grammar Interface
incongruent responses to both agent-focused and patient-focused trials, except for the 3Year-Olds with patient-focused trials.
Discussion
We set out to investigate whether children can use the active passive alternation –
essentially a choice of subject/focus – in a way that is consistent with the eye-gaze of the
speaker – a task that requires integrating social and syntactic knowledge. We established
that 4-Year-Olds and adults are able to use speaker-gaze to choose a felicitous subject when
describing a scene; 3-Year-Olds can do so consistently only when the focus is on the agent.
Our interpretation of how participants achieve this performance and of the relatively
impaired performance of the 3-Year-Olds on the passive follows. To succeed at this task is
not trivial. We don’t know for sure all the steps needed to produce the appropriate response
but at the very least, the following processes are all credibly involved.
First, both children and adults need to understand that following gaze establishes
reference. Subcomponents of this ability are recognizing that looking is intentional behavior
directed to external objects and events; that looking results in the mental experience of
seeing an object or event; and that others share in the capacity to see things (D'Entremont, et.
al., 2007). Second, participants need to coordinate where their attention has been focused
with a linguistic representation. This involves selecting the construction that best serves the
function of foregrounding a participant, which in this case is the subject position (perhaps
something like Fig. 1). Part of this ability requires suppressing the preferred information
structure and most heavily entrenched form (the active) when one needs to describe a scene
from the perspective of the patient. There is evidence that this maybe more of a challenge
for the 3-Year-Olds because they have the strongest preference for describing the scene with
an active in the absence of social cues (see Control Condition). In addition, success on the
23
The Attention-Grammar Interface
experiment not only requires attention for a given trial but the ability to switch attention
between trials. Thus inhibitory control, attentional flexibility, and working memory are all
implicated in giving the correct response, which as noted in the introduction, are still in
development around 4 to 5 years-of-age (Hughes, 1998). Finally, participants need to
produce a string of nouns, verbs and auxiliaries that not only satisfy the grammatical
requirements of who did what to whom but also conform to the social-pragmatic demands of
the context.
Salamo, Lieven and Tomasello (2010) note that children have more difficulty giving
pragmatically appropriate responses to sentence-focus questions of the type “what’s
happening?” than either argument-focus questions “who is VERB-ing?” or predicate-focus
questions “what is AGENT doing?” This pattern corresponds with sentence-focus questions
being relatively less frequent to the other types in child directed speech. In this experiment
we deliberately chose an unbiased question (linguistically-speaking) to assess the role of
social cueing in isolation (i.e., “what’s happening?”). To the extent that sentence-focus
questions are more pragmatically difficult, this seems to affect the youngest age group in
this study the most. Again, one possible explanation for this is in terms of the development
of underlying domain-general capabilities – the idea is that argument-focus questions and
predicate-focus questions help to anchor the relevant piece of information (either verb or
agent) in short term-memory, from which the appropriate response to the question is
constructed (an advantage which is not present in sentence-focus questions). Although more
infrequent in child directed speech, sentence-focus questions may benefit the most from the
support of social cues precisely because they are linguistically uninformative. Indeed Nappa
et al., (2009) found the strongest effect of social cues on verb learning in the linguistically
uninformative condition. In reality children face a complex matrix of social and linguistic
24
The Attention-Grammar Interface
cues and they have to work out how reliable these are across multiple situations. Using the
design in this study, one might predict an even stronger effect of subject choice alternations
(and perhaps at younger age) where linguistic anchoring (e.g., “what’s happening to the
cat?”) acts in coalition with social cues (e.g., speaker looks at cat).
As noted in the introduction, in English the active is the preferred information
structure for describing a causal transitive scene: actives outnumber passives 9:1 in written
text and the ratios are even higher for spoken discourse. Not only this, the token frequency
for passives is low. In one analysis of child directed speech, the passive-per-utterance rate is
0.36% (Gordon and Chafetz, 1990). Thus all age groups have had relatively less experience
with the passive than the active, and the results show that there are more congruent active
responses than passive responses for all age groups. The active construction also aligns with
a non-linguistic preference for conceptually framing events as source-to-goal or instigatorpatient rather than the reverse (Lakusta and Landau, 2005; Lakusta, Wagner, O’Hearn &
Landau, 2007). Nappa et al. (2009) note that in many languages, verbs that align with this
preference (e.g., chase) outnumber their asymmetric partners (e.g., flee). They also note the
instigator-patient bias is present pre-linguistically (Csibra, Biro, Koos, & Gergely, 2003;
Woodward, 1998) and continues through to adulthood (Gleitman et al., 2005). Increased
experience (use) of the passive with time, particularly with a literate undergraduate
population from which the adult sample was mainly drawn (see Dąbrowska, 2006), acts in
opposition to this tendency. This experience would make it relatively easier to retrieve the
passive construction from memory when it is required (i.e., on the patient-focused trials).
Also, in the design of this experiment, all participants were explicitly shown in the training
phase (and perhaps reminded in the case of adults) that actives and passives were legitimate
ways to describe the scene. This part of the design may have increased the conceptual
25
The Attention-Grammar Interface
accessibility (Osgood & Bock, 1977) of the passive and boosted the level of passives
description from that which we would normally see in spontaneous speech.
Experimentally it has been shown children around their second year of life can use
the attentional and emotional states of speakers as cue to their communicative intentions
(e.g., Baldwin, 1991; Tomasello and Farrar, 1986). One might wonder why it takes
participants in this study until their fifth year of life to link the linguistic and the social
domains in a way that allows them to succeed on both agent and patient focused trials. As
discussed above, the complexity of the task and the relative frequency of the passive
construction hints at one answer but another possibility, as mentioned in the introduction, is
that it takes time for young children to integrate eye-gaze information (as a basic perceptual
cue) into a more sophisticated picture of how other people work including their future
intentions and mental states (Baron-Cohen, 1994; see Striano & Reid, 2006, for a recent
review). The difference between the 3 and 4 year olds in our experiment might also be
partly attributable to development in this social-cognitive capacity. Without further tests
that can disassociate the lower-level perceptual cues from intentional eye-gaze cues (for
example a non-voluntary eye-gaze shift) it is not clear to what extent the children in this
experiment are using in the intentions of the speaker to assign a subject. For that reason we
use the term the attention-grammar interface rather than the social-grammar interface.
It is also worth noting that the experiments that have focused on the relationship
between attentional states of speakers and communicative intentions (e.g., Baldwin, 1991;
Tomasello and Farrar, 1986) have mainly focused of word learning. The adult-like function
of subjecthood makes it a much more abstract and less concrete learning challenge than
learning words, which, in the usage-based framework at least, means that mastery of
subjecthood requires more evidence and more experience with using it. While the 3-year26
The Attention-Grammar Interface
olds in this study probably possess many of the social-cognitive foundations that the 4-yearolds do, success on the task needs competence in linguistic and executive control domains
as well as effective connections between these domains.
Priming studies show that children already have some representation of the passive
younger than 4 years-of-age (e.g., Huttenlocher, Vasilyeva & Shimpi, 2004). One might
expect that the 3-Year-Olds in this study might be able to succeed on both agent-focused
and patient-focused trials. However, the factors we have reviewed above – the steps needed
to succeed at this task, the relative frequency of the passive, the conceptual bias for
instigator-patient, the relative difficulty with sentence-focus questions, the developing
executive control, the abstractness of ‘subject’ relative to word learning – could all
disproportionally affect the youngest age group.
The performance of the 3-, 4-year-olds and the adults in this study provides further
support for the subject-is-theme-is-attention attention hypothesis (Tomlin 1995; 1997).
What is more, we have demonstrated this using a more ecologically valid cue than a red
arrow hovering above a participant, namely, eye-gaze. Our interpretation of these results is
that eye-gaze plays a similar role to the flashing arrow in the Tomlin studies, namely, it is an
attention-directing cue that foregrounds one character and, by definition, backgrounds the
other. Importantly, we know adults are more likely to look at the subject of their sentence
than any other character (Gleitman, January, Nappa & Trueswell, 2007; Griffin & Bock,
2000). This raises the possibility that young children could be using the social cue of eyegaze in situ (which directs attention) to infer the function of grammatical subject (which is
grounded in attention).
More generally, the methodology we have used here advocates exploring linguistic
cues in combination with the social-pragmatic context. By using eye-gaze we have been
27
The Attention-Grammar Interface
able to consider a broader range of cues than a traditional corpus-based approach to the
development of language. By doing so, we have been able to get closer to reconstructing the
rich social-pragmatic-linguistic world in which the child actually grows up. The challenge is
to explore ways in which social-pragmatic skills interact with prodigious pattern-finding
abilities in a way that which explains the emergence of linguistic knowledge. Yu & Ballard
(2007) found that a computational model of word-learning performed better when it used
social information (joint attention and prosody) in combination with statistical cues (crosssituational learning) than when it relied on purely statistical information alone. There is no
reason to doubt that a combination of social and linguistic cues also helps in learning
syntactic relations (e.g., Nappa et al., 2009).
We noted in the introduction that, typically, children do not produce full passives in
spontaneous speech until about 4-5 years-of-age however performance can be significantly
boosted when the passive is supported with case marked pronouns (Ibbotson et al., 2009)
with training (Pinker, Lebeaux & Frost 1987; Brooks & Tomasello, 1999) and when the
passive form is more frequent in the ambient language (relative to Indo-European languages)
(Allen & Crago, 1996; Suzman, 1985; Pye & Quixtan Poz, 1988). To this list we might add
that felicitous passive use is also boosted when supported with the kind of social-cognitive
cues that children might actually experience, such as eye-gaze.
Our understanding of children’s social cognition has progressed significantly,
however, it has yet to be worked through in any detail how this knowledge interacts with
emerging syntactic representations. This paper is a step in that direction and investigates
whether eye-gaze influences the choice of grammatical subject for young children and
adults. The linguistic notion of subjecthood is in principle an abstract one. Previous work
suggests the function of the subject position can be grounded in terms of attention and
28
The Attention-Grammar Interface
information structure. One powerful source of directing attention that we know young
children are sensitive to is eye-gaze. We have demonstrated that 4-year-olds and adults (but
not conclusively 3-year-olds) are able to use speaker-gaze to choose a felicitous subject. It
has been shown before that social-cognitive cues help children learn words but this is the
first demonstration that eye-gaze could be important in learning something as abstract as
subject role. Integrating attentional and grammatical information in this way allows children
to limit the degrees of freedom on what the function of certain linguistic constructions might
be (cf. Nappa et al., 2009) and allows linguistic theory to ground abstract functions in
deeper cognitive and communicative principles (Tomlin, 1995,1997; Goldberg, 2006;
Langacker, 1991).
29
The Attention-Grammar Interface
References
Allen, S. & Crago, M. (1996). Early passive acquisition in Inuktitut. Journal of Child
Language 23(1). 129-155.
Ando, S. (2002). Luminance-induced shift in the apparent direction of gaze. Perception, 31,
657–674.
Baldwin, D. A. (1991). Infant contribution to the achievement of joint reference. Child
Development, 62, 875–890.
Baron-Cohen, S. (1994). How to build a baby that can read minds: Cognitive mechanisms in
mindreading. Cahiers de Psychologie Cognitive [Current Psychology of Cognition], 13,
513–552.
Baron-Cohen, S., Campbell, R., Karmiloff-Smith, R., Grant, J., & Walker, J. (1995). Are
children with autism blind to the mentalistic significance of the eyes? British Journal of
Developmental Psychology, 13, 379–398.
Braine, M.D.S., Brooks, P.J., Cowan, N., Samuels, M.C., & Tamis-LeMonda, C. (1993).
The Development of categories at the semantics/syntax interface. Cognitive Development, 8,
465-494.
Brooks, P. J. and Tomasello, M. (1999). Young children learn to produce passives with
nonce verbs. Developmental Psychology, 35(1), 29-44.
Burger, J., Gochfeld, M., & Murray, B. G. (1992). Risk discrimination of eye contact and
directness of approach in black iguanas (Ctenosaura similis). Journal of Comparative
Psychology, 106, 97–101.
30
The Attention-Grammar Interface
Burghardt, G. M., & Greene, H. W. (1990). Predator simulation and duration of death
feigning in neonate hognose snakes. Animal Behaviour, 36, 1842–1844.
Campbell, R., Heywood, C. A., Cowey, A., Regard, M., & Landis, T. (1990). Sensitivity to
eye gaze in prosopagnosic patients and monkeys with superior temporal sulcus ablation.
Neuropsychologia, 28, 1123–1142.
Carlson, S. M. (2005). Developmentally sensitive measures of executive function in
preschool children. Developmental Neuropsychology, 28, 595-616.
Carpenter, M., Akhtar, N., & Tomasello, M. (1998). Fourteen- through 18-month-old infants
differentially imitate intentional and accidental actions. Infant Behavior and Development,
21(2), 315–330.
Charman, T., Baron-Cohen, S., Swettenham, J., Baird, G., Cox, A., & Drew, A. (2001).
Testing joint attention, imitation, and play as infancy precursors to language and theory of
mind. Cognitive Development, 15, 481–498.
Choi, S., & Bowerman, M. (1991). Learning to express motion events in English and
Korean: The influence of language-specific lexicalization patterns. Cognition, 43, 83–121.
Corkum, V., Byrne, J. M., & Ellsworth, C. (1995). Clinical assessment of sustained
attention in preschoolers. Child Neuropsychology, 1, 3–18.
Corrigan, R. (1988). Children's identification of actors and patients in prototypical and nonprototypical sentence types. Cognitive Development, 3, 285-297.
Croft, W. (2001). Radical Construction Grammar: syntactic theory in typological
perspective. Oxford: Oxford University Press.
31
The Attention-Grammar Interface
Csibra, G., Biro, S., Koos, O., & Gergely, G. (2003). One year old infants use teleological
representations of actions productively. Cognitive Science, 27, 111–133.
Dąbrowska, E. and Street, J. (2006). Individual differences in language attainment:
Comprehension of passive sentences by native and non-native English speakers. Language
Sciences 28, 604-615.
Day, M. C. (1978). Visual search by children: The effect of background variation and the
use of visual cues. Journal of Experimental Child Psychology, 25, 1–16.
D'Entremont, B., Yazbeck, A., Morgan, A. & MacAulay, S. (2007). Early gaze-following
and the understanding of others. In R. Flomm, K. Lee & D. Muir (Eds.), Gaze-following:
It's development and significance (pp. 77–94). Mahwah, NJ: Lawrence Erlbaum Associates.
Enns. J. T., & Girgus, J. S. (1985). Developmental changes in selective and integrative
visual attention. Journal of Experimental Child Psychology, 40, 319-337.
Erickson, T. A. and Mattson, M. E. (1981). From words to meaning: A semantic illusion.
Journal of Verbal Learning and Verbal Behavior, 20: 540-552.
Farroni, T., Csibra, G., Simion, F., & Johnson, M. H. (2002). Eye contact detection in
humans from birth. Proceedings of the National Academy of Sciences, 99, 9602–9605.
Farroni, T., Johnson, M. H., & Csibra, G. (2004). Mechanisms of eye gaze perception
during infancy. Journal of Cognitive Neuroscience, 16, 1320–1326.
Farroni, T., Massaccesi, S., Menon, E., & Johnson, M. H. (2007). Direct gaze modulates
face recognition in young infants. Cognition, 102, 396–404.
32
The Attention-Grammar Interface
Farroni, T., Massaccesi, S., Pividori, D., & Johnson, M. H. (2004). Gaze following in
newborns. Infancy, 5, 39–60.
Fillmore, C. (1976). Frame Semantics and the Nature of Language. In "Annals of the New
York Academy of Sciences: Conference on the Origin and Development of Language and
Speech". Vol. 280, 20-32.
Friere, A., Eskritt, M., & Lee, K. (2004). Are eyes windows to a deceiver’s soul? Children’s
use of another’s eye gaze cues in a deceptive situation. Developmental Psychology, 40,
1093–1104.
ergely, ., ekkering, H., & Kir ly, I. (2002). Rational imitation in preverbal infants.
Nature, 415, 755.
Gibson, E. J., & Yonas, A. (1966). A developmental study of the effects of visual and
auditory interference on a visual scanning task. Psychonomic Science, 5, 163–164.
Givon, T. 1992. The grammar of referential coherence as mental processing instructions.
Linguistics 30. 5–55.
Gleitman, L. R., January, D., Nappa, R., & Trueswell J. C. (2007). On the give and take
between event apprehension and utterance formulation. Journal of Memory and Language,
57(4), 544–569.
Gleitman, L., Cassidy, K., Nappa, R., Papafragou, A., & Trueswell, J. (2005). Hard words.
Language Learning and Development, 1(1), 23–64.
Goldberg, A. (2006). Constructions at Work: The nature of generalisations in language.
Oxford: Oxford University Press.
33
The Attention-Grammar Interface
Gordon, P. & J. Chafetz. (1990). Verb-based versus class-based accounts of actionality
effects in children’s comprehension of passives. Cognition, 36, 227-254.
Griffin, Z. M., & Bock, K. (2000). What the eyes say about speaking. Psychological Science,
11, 274–279.
Guttentag, R. E. (1985). Memory and aging: Implications for theories of memory
development during childhood. Developmental Review, 5, 56-82.
Halliday, M. A. K. (1985). An introduction to functional grammar. London: Edward Arnold.
Henderson, J., and F. Ferreira. 2004. The integration o and f language, vision, and action:
eye movements and the visual world. New York: Psychology Press.
Hood, B. M., Macrae, C. N., Cole-Davies, V., & Dias, M. (2003). Eye remember you: The
effects of gaze direction on face recognition in children and adults. Developmental Science,
6, 67–71.
Hughes, C. (1998). Executive function in preschoolers: Links with theory of mind and
verbal ability. British Journal of Developmental Psychology, 16, 233-253
Huttenlocher, J., Vasilyeva, M. & Shimpi, P. (2004). Syntactic priming in young children.
Journal of Memory and Language, 50, 182-195.
Hwang, H., and E. Kaiser. (2009). The effects of lexical vs. perceptual primes on sentence
production in Korean: an on-line investigation of event apprehension and sentence
formulation. Talk presented at the 22nd CUNY conference on sentence processing, Davis,
CA.
34
The Attention-Grammar Interface
Ibbotson, P. (2010). A Cognitive Constructivist Approach to Early Syntax Acquisition. PhD
thesis (unpublished), submitted to the University of Manchester.
Ibbotson, P., Theakston, A., Lieven, E., & Tomasello, M. (2011). The Role of Pronoun
Frames in Early Comprehension of Transitive Constructions in English. Language, Learning
and Development. 7: 24–39.
Jackendoff, R. (1983). Semantics and cognition. Cambridge, MA: MIT Press.
Kannass, K. N., & Oakes, L. M. (2008). The development of attention and its relations to
cognitive skills in infancy and toddlerhood. Journal of Cognition and Development, 9, 222–
246.
Keenan, E. (1976). Towards a universal definition of “subject”. C.N. Li (ed.), Subject and
topic, 303 – 33. New York: Academic Press.
Kobayashi, H., & Kohshima, S. (1997, June 19). Unique morphology of the human eye.
Nature, 387, 767–768.
Lakoff, G. (1987). Cognitive models and prototype theory. In U. Neisser (Ed.), Concepts
and conceptual development: Ecological and intellectual factors in categorization. Emory
symposia in cognition (Vol. 1, pp. 63–100). Cambridge, England: Cambridge University
Press.
Lakusta, L, Wagner, L., O’Hearn, K., & Landau, . (2007) Conceptual foundations of
spatial language: Evidence for a goal bias in infants. Language Learning and Development,
3(3), 179–197.
Lakusta, L., & Landau, B. (2005). Starting at the end: The importance of goals in spatial
language. Cognition, 96, 1–33.
35
The Attention-Grammar Interface
Landau, ., and R. Jackendoff. 1993. ‘‘What’’ and ‘‘where’’ in spatial language and spatial
cognition. Behavioral and Brain Sciences 16. 217–65.
Langacker, R.W. (1991). Foundations of cognitive grammar: Vol. 2–Descriptive application.
Stanford: Stanford University Press.
Langacker, Ronald (1987). Foundations of cognitive grammar: Theoretical prerequisites
(vol. 1). Stanford: Stanford University Press.
Langacker, Ronald 1991. Foundations of Cognitive Grammar, vol. 2.: Descriptive
Application. Stanford: Stanford University Press.
Langacker, Ronald W. (1988). Autonomy, agreement, and cognitive grammar. Agreement
in Grammatical Theory, ed. by Diane Brentari, Gary Larson, and Lynn MacLeod, 147-180.
Chicago: Chicago Linguistic Society.
Lee, K., Eskritt, M., Symons, L. A., & Muir, D. (1998). Children’s use of eye gaze for
“mind reading”. Developmental Psychology, 34, 525–539. MacWhinney, B. (1977). Starting
points. Language, 53, 152-168.
MacWhinney, B. 1977. Starting points. Language 53(1). 152–68.
Morales, M., Mundy, P., & Rojas, J. (1998). Following the direction of gaze and language
development in 6-month-olds. Infant Behavior and Development, 21, 373–377.
Morales, M., Mundy, P., Delgado, C. E. F., Yale, M., Neal, A. R., & Schwartz, H. K. (2000).
Gaze following, temperament, and language development in 6-month-olds: A replication
and extension. Infant Behavior and Development, 23, 231–236.
36
The Attention-Grammar Interface
Mumme, D., Bushnell, E., DiCorcia, J. & Lariviere, L. (2007). Infants' use of gaze cues to
interpret others' actions and emotional reactions. In R. Flomm, K. Lee & D. Muir (Eds.),
Gaze-following: It's development and significance (pp. 143–170). Mahwah, NJ: Lawrence
Erlbaum Associates
Myachykov, A, and R. S. Tomlin. (2008). Perceptual priming and structural choice in
Russian sentence production. Journal of Cognitive Science 6(1). 31–48.
Myachykov, A, S. Garrod, and C. Scheepers. (2010). Perceptual priming of structural choice
during English and Finnish sentence production. Language & Cognition: state of the art, ed.
by R. K. Mishra and N. Srinivasan,54–72. Munich: Lincom Europa.
Myachykov, A., Thompson, D., Scheepers, C., & Garrod, S. (2011). Visual attention and
structural choice in sentence production across languages. Language and Linguistic
Compass, 5(2), 95-107. DOI: 10.1111/j.1749-818X.2010.00265.x
Nappa, R., Wessell, A., McEldoon, K.L., Gleitman, L.R. & Trueswell, J.C. (2009). Use of
Speaker’s aze and Syntax in Verb Learning. Language Learning and Development, 5(4),
203-234.
Osgood, C. E., and ock, J. K. (1977). “Salience and sentencing: some production
principles,” in Sentence Production: Developments in Research and Theory, ed. S.
Rosenberg (Hills- dale, NJ: Erlbaum), 89–140.
Papafragou, A., Cassidy, K., & Gleitman, L. (2007). When we think about thinking: The
acquisition of belief verbs. Cognition, 105, 125–165.
Pellicano, E., & Rhodes, G. (2003). The role of eye-gaze in understanding other minds.
British Journal of Developmental Psychology, 21, 33–43.
37
The Attention-Grammar Interface
Perrett, D. I., & Mistlin, A. J. (1991). Perception of facial characteristics by monkeys. In W.
C. Stebbins & M. A. Berkley (Eds.), Comparative perception (pp. 187–215). New York:
Wiley.
Pick, A. D., and Frankel, G. W. (1974) A developmental study of strategies of visual
selectivity. Child Dev, 45: 1162-1165.
Pinker, S., Lebeaux, D. and Frost, L. (1987). Productivity and constraints in the acquisition
of the passive. Cognition 26. 267.
Posner, M. I. 1980. Orienting of attention. Quarterly Journal of Experimental Psychology 32.
3–25.
Poulin-Dubois, P., & Forbes, J. (2002). Toddlers’ attention to intentions-in-action in
learning novel action words. Developmental Psychology, 38, 104–114.
Prentice, J. L. 1967. Effects of cuing actor vs. cuing object on word order in sentence
production. Psychonomic Science 8. 163–4.
Pye, C. & Poz, P. . (1988). Precocious passives (and antipassives) in uich Mayan.
Papers and Reports on Child Language Development 27. 71-80.
Radford, A. et al. (1999) Linguistics: an introduction. Cambridge:CUP.
Rispoli, M. (1991). The mosaic acquisition of grammatical relations. Journal of Child
Language, 18, 517-551.
Ruff, H. A., & Rothbart, M. K. (1996). Attention in early development: Themes and
variations. New York: Oxford University Press.
38
The Attention-Grammar Interface
Salomo, D., Lieven, E., & Tomasello, M. (2010). Young children's sensitivity to new and
given information when answering predicate-focus questions. Applied Psycholinguistics,
31(1), 101-115.
Striano, T., & Reid, V. M. (2006). Social cognition in the first year. Trends in Cognitive
Sciences, 10, 471–476.
Suzman, S. (1985). Learning the passive in Zulu. Papers and Reports on Child Language
Development 24. 131-37.
Symons, L. A., Hains, S. M. J., & Muir, D. W. (1998). Look at me: Five-month-old infants’
sensitivity to very small deviations in eye-gaze during social interactions. Infant Behavior
and Development, 21, 531–536.
Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science, 12, 49–
100.
Talmy, L. (2000). Toward a cognitive semantics. Cambridge, MA: MIT Press.
Talmy, Leonard. (1985). Lexicalization patterns: semantic structure in lexical forms. In T.
Shopen (ed.), Language typology and syntactic description III, 57-149. New York:
Cambridge University Press.
Talmy,L. (2007). Attention phenomena. In Dirk Geeraerts & Hubert Cuyckens (eds.).
Oxford Handbook of Cognitive Linguistics Oxford University Press, 264-293.
Taylor, J. (2002). Cognitive grammar. Oxford, England: Oxford University Press.
Tomasello, M. (2003). Constructing a Language: A Usage-Based Theory of Language
Acquisition. Harvard University Press.
39
The Attention-Grammar Interface
Tomasello, M., & Barton, M. (1994). Learning words in nonostensive contexts.
Developmental Psychology, 30, 639–650.
Tomasello, M., & Ferrar M. J. (1986). Joint attention and early language. Child
Development, 57, 1454–1463.
Tomlin, R. S. (1995). Focal attention, voice, and word order: an experimental, crosslinguistic study. In P. Downing & M. Noonan (Eds.), Word order in discourse (pp. 517-554).
Amsterdam: John Benjamins.
Tomlin, R. S. (1997). Mapping conceptual representations into linguistic representations:
The role of attention in grammar. In J. Nuyts & E. Pederson (Eds.), Language and
conceptualization (pp. 162-189). Cambridge: Cambridge University Press.
Traxler, Matthew. 2011. Introduction to Psycholinguistics: Understanding Language
Science. Wiley-Blackwell: Chicester, UK.
Trueswell, J. C., and M. K. Tanenhaus. 2005. Approaches to studying world-situated
language use. Cambridge, MA: The MIT Press.
Turner, E. A., and R. Rommetveit. 1968. Focus of attention in recall of active and passive
sentences. Journal of Verbal Learning and Verbal Behavior 7(2). 543–8.
Well, A.D., Larch, E.P., & Anderson, D.R. (1980). Developmental trends in distractibility:
Is absolute or proportional decrement the appropriate measure of interference? Journal of
Experimental Child Psychology, 30, 109-124.
Woodward, A. L. (1998). Infants selectively encode the goal object of an actor’s reach.
Cognition, 69, 1–34.
40
The Attention-Grammar Interface
Woodward, A. L., & Sommerville, J. A. (2000). Twelve-month-old infants interpret action
in context. Psychological Science, 11, 73–77.
Yu, C. & Ballard, D. (2007). A unified model of early word learning: Integrating statistical
and social cues. Neurocomputing 70, 2149-2165.
41
The Attention-Grammar Interface
Acknowledgements
Thank you to St Vincent’s Infant School, Freshfields Nursery and all the children who took
part in this study; to Anna Coates and Mickie Glover for also assisting in this experiment; to
Andriy Myachykov, Ewa Dabrowska and one anonymous reviewer who provided helpful
comments on the manuscript. This research was supported by a postdoctoral fellowship
from the Max Planck Institute for Evolutionary Anthropology, Leipzig.
42